1. Field of the Invention
The present invention relates to a technique of dealing with multiple link failures occurring in a communication network.
2. Description of the Related Art
A transmission control procedure called a ring protocol is generally adopted in topology of a ring structure (hereafter referred to as ring topology) in an Ethernet (registered trademark) communication network. The ring protocol is a layer 2 network redundant protocol. Multiple devices relaying user frames in a ring (hereafter these devices are referred to as relay devices) are interlinked in the ring structure. At least one of the relay devices is assigned to a device sending a frame (monitor frame) for checking the state of the ring (hereafter this device is referred to as monitor device). This arrangement of the ring protocol continuously monitors the state of the ring and controls ports of the respective devices to detect the occurrence of any failure and immediately change the communication route in response to the detection. The ring protocol enables the prompt change of the communication route in the event of the occurrence of a failure. Another advantage of the ring topology reduces the total number of transmission paths and interfaces, compared with mesh topology of connecting the respective relay devices in a net structure.
Since one single ring can not deal with plural failures occurring in the network, the ring protocol generally adopts a structure of using multiple rings interlinked via a linkage (hereafter referred to as common link) to ensure the redundancy (hereafter referred to as multi-ring structure with a common link).
In the multi-ring structure with a common link, however, in the event of a failure occurring on the common link, the interlinked rings may cause a large loop (super-loop). Various systems have been proposed to solve this problem in the ring protocol.
One proposed system controls a device located at a terminal end of a common link (hereafter this device is referred to as common link terminal device) shared by two interlinked rings to send an auxiliary monitor frame to a monitor device of one ring (see Japanese Patent Laid-Open No. 2004-201009). This assigns the ring to a non-working ring of monitoring the occurrence of a failure on the common link. Even in the event of some failure on the common link, the monitor device receiving the auxiliary monitor frame (hereafter this device is referred to as auxiliary monitor device) keeps one of ports blocked to allow transmission of monitor frames but prohibit transmission of user frames, with a view to avoid the occurrence of the super-loop.
Another proposed system sets a preferential order to multiple rings and, in the event of a failure occurring on a common link shared by the multiple rings, causes a monitor frame in a ring of the higher priority to be introduced to a ring of the lower priority. Circling the monitor frame in these rings controls the blocking (see ‘Foundry Biglron RX Series Configuration Guide’, Chapter 15, Metro Ring Protocol (MRP)).
The multi-ring structure with the common link can deal with plural failures with some restriction in the ring topology. In the prior art system of the former cited reference, however, in a multi-failure state with one failure on the common link and another failure on another link in the ring assigned to the working ring of monitoring the common link, the auxiliary monitor frame sent from the common link terminal device reaches the monitor device in the ring assigned to the non-working ring of monitoring the common link. The reception of the auxiliary monitor frame interferes with release of the blocked port in the monitor device. This leads to failed recovery of communication even in the presence of the physical connection.
In the prior art system of the latter cited reference, on the other hand, in a multi-failure state with one failure on the common link and another failure on another link in the ring assigned to the working ring of monitoring the common link, the monitor frame does not reach the monitor device in the ring assigned to the non-working ring of monitoring the common link. This allows release of the blocked port in the monitor device and is free from the above problem arising in the system of the former cited reference. The system of the latter cited reference, however, still has some disadvantages. The occurrence of each failure on the common link increases the number of relay devices as passing points of a monitor frame. This causes the delay of the monitor frame and increases the potential for wasting the frame due to the convergence.